sign in sign up
<<
Home , Carbamate kinase, Carbamoyl phosphate, Ligase

Proc. Natl. Acad. Sci. USA Vol. 94, pp. 12348–12353, November 1997

Novel mechanism for carbamoyl- synthetase: A switch for functionally equivalent domains

MICHAEL KOTHE*, BINNUR EROGLU*, HOLLY MAZZA,HENRY SAMUDERA, AND SUSAN POWERS-LEE†

Department of Biology, Northeastern University, Boston, MA 02115

Communicated by Robert H. Abeles, Brandeis University, Waltham, MA, August 26, 1997 (received for review July 7, 1997)

ABSTRACT Carbamoyl-phosphate synthetases (CPSs) uti- modeling seemed appropriate for the following reasons. First, lize two molecules of ATP at two internally duplicated domains, B there is significant sequence identity shared by portions of BN (as and C. Domains B and C have recently been shown to be well as other -dependent ) and each of the CPS structurally [Thoden, J. B., Holden, H. M., Wesenberg, G., domains B and C (2). Also, both CPS and BN couple cleavage of Biochemistry Ϫ Raushel, F. M. & Rayment, I. (1997) 36, 6305–6316] ATP to ADP with the activation of HCO3 to yield CxP, and both and functionally [Guy, H. I. & Evans, D. R. (1996) J. Biol. Chem. catalyze reaction of the activated carboxyl group with an amino 271, 13762–13769] equivalent. We have carried out a site-directed group (NH3 for CPS and the N1Ј group of biotin for BN; ref. 25). mutagenic analysis that is consistent with ATP binding to a A final rationale for the homology modeling was that the BN palmate motif rather than to a Walker A͞B motif in domains B structure appears to represent a general fold for enzymes that and C. To accommodate our present findings, as well as the other catalyze formation of an amide bond accompanied by cleavage of recent findings of structural and functional equivalence, we are ATP to ADP and formation of an acyl-phosphate intermediate proposing a novel mechanism for CPS. In this mechanism utili- (26). The ATP within this structure is comprised of a zation of ATP bound to domain C is coupled to carbamoyl- ‘‘palmate’’ motif (27) rather than the Walker A͞B motif (4). Other phosphate synthesis at domain B via a nucleotide switch, with the members of this N- structural group (enzymes catalyzing energy of ATP hydrolysis at domain C allowing domain B to cycle between two alternative conformations. amide ligation and sharing structural folding patterns) are gluta- thione synthetase (GS; ref. 27), which catalyzes phosphorylation of a ␥-glutamyl-cysteinyl carboxyl group and ligation to the amino Carbamoyl-phosphate (CP) is a high-energy biological com- group of glycine, and D-alanine:D-alanine ligase (AA; ref. 28), pound that plays a key role in the introduction of both which catalyzes phosphorylation of one D-alanyl carboxyl group and single carbon units into the metabolic pool. CP may serve either as the precursor for synthesis of or, and ligation to the amino group of the other molecule of D-alanine. in an alternative pathway, as the precursor for . In Although our homology modeling served as the basis for the other addition to serving as a required component of almost all studies described here, reliance solely on it has been rendered proteins, arginine is used in the liver for ammonia detoxification unnecessary by the recent report of the x-ray crystal structure for via the cycle and, under appropriate physiological condi- E. coli CPS (29). Domains B and C were found to superimpose tions, it is also used in a variety of tissues for nitric oxide with an rmsd of 1.1 Å and superposition of domain B on the BN formation. Depending on the physiological role, CP synthetases coordinates yielded a rms value of 1.7 Å. (CPSs) vary in mode of regulation and subunit composition (1). We have utilized site-directed mutagenesis to determine that However, all of the CPSs thus far sequenced display marked appropriate CPS residues display the functional characteristics primary sequence identity (2, 3), and all catalyze the formation of expected for ATP binding to a palmate structural motif (27) but Ϫ CP, Pi, and two ADP from HCO3 ,NH3, and two ATP. not those expected for ATP binding to potential Walker A͞B Sequence analysis demonstrated regions of internal duplication structural motifs (4). The fact that domains B and C are struc- within CPS and suggested, on the basis of similarities to the turally equivalent suggested to us that they should display at least ATP-binding sequence motifs of Walker A͞B proteins (i.e., pro- quasi-equivalent utilization of the two ATP molecules. This teins involving a ␤␣␤ nucleotide binding fold; ref. 4), that each hypothesis is also consistent with a recent report demonstrating region of internal duplication contains one ATP site (5, 6). potential functional equivalence of domains B and C (30); in Subsequent studies (7–10) demonstrated that the synthetase du- surprising contrast to the distinct functional behavior of domains plications correspond to independently folded domains (B and C) B and C within the intact CPS, Guy and Evans (30) found that and that binding of one ATP molecule is localized to each of the CPS from which either domain B or C has been deleted can two synthetase domains (6, 11–13). A great deal of experimental dimerize to display full functional ability. evidence (14–23) has established that the role of one ATP is to However, neither this study nor our own present findings are form the -bound intermediate carboxy-phosphate (CxP) consistent with the sequential reaction mechanism (Fig. 1) for and that domain B binds this ATP (ATPB) whereas domain C CPS first proposed by Jones and Lipmann in 1960 (31), and still binds the other ATP molecule (ATP ). C generally utilized as the working mechanistic model for CPS (1, When we carried out the presently described studies, further 30, 32, 33). As noted above, CxP has been demonstrated to be definition of the utilization of ATP and ATP by CPS could not B C an intermediate (14–21), establishing the validity of Eq. 1. be based on a solved three-dimensional structure because there was none available for any CPS at that time. However, we found Although the remainder of the proposed pathway has not been that domains B and C of CPS could each be fit to the structural established experimentally, it has been generally assumed that coordinates for the (BN) component of Esch- the CPS reaction proceeds sequentially from Eq. 1: CxP erichia coli acetyl Co-A carboxylase (24). This BN homology undergoes nucleophilic attack by NH3 to yield a tetrahedral

The publication costs of this article were defrayed in part by page charge Abbreviations: AA, D-alanine:D-alanine ligase; BN, biotin carboxylase; CP, carbamoyl-phosphate; CPS, carbamoyl-phosphate synthetase; CxP, payment. This article must therefore be hereby marked ‘‘advertisement’’ in carboxy-phosphate; N-ligase group, enzymes catalyzing amide ligation accordance with 18 U.S.C. §1734 solely to indicate this fact. and sharing structural folding pattern; GS, . © 1997 by The National Academy of Sciences 0027-8424͞97͞9412348-6$2.00͞0 *M.K. and B.E. contributed equally to this study. PNAS is available online at http:͞͞www.pnas.org. †To whom reprint requests should be addressed. e-mail: [email protected].

12348 Downloaded by guest on September 25, 2021 Biochemistry: Kothe et al. Proc. Natl. Acad. Sci. USA 94 (1997) 12349

FIG. 1. Sequential mechanism for CPS (31).

intermediate (Eq. 2); the tetrahedral intermediate breaks x-ray coordinates are available (June 1998 is listed by the Protein down to yield carbamate plus Pi (Eq. 3); and finally carbamate Data Bank as the estimated date of release for the E. coli CPS interacts with the second molecule of ATP in a carbamate coordinates). However, the passage of the hydrophobic (and reaction to yield CP (Eq. 4). chemically stable) indole group through the well-defined 25-Å We are proposing an alternative mechanism for CPS (Fig. 2, hydrophobic tunnel of tryptophan may not be an ap- discussed in detail below) in which usage of the two molecules of propriate model for possible movement of the labile and polar ATP is coupled, with the first ATP bound (ATPC in intact CPS) carbamate intermediate. And, as discussed below, the spatially acting as a molecular switch to trigger conformational changes in distinct active sites would be consistent with our proposed cou- domain B. Domain B can then catalyze the formation of CP via pled reaction mechanism (Fig. 2). Eqs. 1, 2, and 3૾, and finally of Eqs. 1, 2, and 3 on domain C allows return to the initial CPS conformation and the release MATERIALS AND METHODS of products. Although many of the features of the original Molecular biological procedures were performed essentially as scheme, including its pioneering proposal of energy transfer via described in Sambrook et al. (36). Standard yeast genetic and the ␥-phosphate group of ATP, are retained in Fig. 2, this revised molecular biological procedures were performed as described scheme also incorporates the more recent knowledge of the role (37). Site-directed mutagenesis was performed by the recombinant of ATP in effecting protein conformational switches (34). PCR method (38) or the unique site-elimination method (39) as The sequential mechanism (Fig. 1) was assumed in interpreting described for yeast arginine-specific CPS (33, 40). The presence of the recent x-ray structure for CPS (29), and the authors did not desired mutations and the absence of any unexpected mutations in consider any mechanistic implications of the finding of structural each mutagenesis cassette was confirmed by sequencing. To equivalence for domains B and C. Strikingly, the apparent ATP determine whether mutant constructions could produce CP, their sites of domains B and C were found to be far removed (with a ability to functionally complement CPS-disrupted yeast strain linear distance of about 35 Å) and it was proposed that, in analogy LPL26 was assessed (41). Generation time determination, crude to (35), the presumed carbamate interme- extract preparation, enzymatic activity assay, protein determina- diate is channeled through the interior of CPS. Detailed evalu- tion, SDS͞PAGE, and Western blot analysis were as previously ation of this channeling proposal will not be possible until the utilized for CPS mutational analysis (33, 40).

FIG. 2. Coupled mechanism for CPS. Downloaded by guest on September 25, 2021 12350 Biochemistry: Kothe et al. Proc. Natl. Acad. Sci. USA 94 (1997)

RESULTS AND DISCUSSION using Molecular Simulations (Waltham, MA) software, including Homology Modeling of Yeast CPS Domains B and C with a BN QUANTA and X-PLOR, on a Silicon Graphics 4D͞35 workstation. Template. As discussed above, sole reliance on our homology The QUANTA SPIN program was used to optimize the van der Waals modeling has been eliminated by the recent report of the x-ray interactions of the substituted side chain atoms. Energy minimi- crystal structure for E. coli CPS (29). However, because the CPS zation was carried out using the CHARMm program (43) to yield the structural coordinates are not yet available, the homology mod- final CPS model structures. The modeled structures for CPS eling is still necessary to provide a three-dimensional structure for domains B and C, as well as a comparison to the BN backbone, are CPS upon which to base interpretation of the site-directed shown in Fig. 4. The backbones of the three structures are generally mutagenesis data. The template BN, as determined by x-ray superimposable, with only small dislocations occurring in a few crystallography (24, 26), is composed of four structural regions: regions of the molecule. Twelve residues are conserved (or con- residues 1–105, 106–203, 204–318, and 319–449. As shown in Fig. servatively substituted) in the ATP binding sites for GS and AA. 3A, portions of the yeast CPS domains B and C (residues 133–343 These 12 residues are also generally conserved in BN and in and 677–887) share sequence identity (20% and 17%, respec- domains B and C of CPS (Fig. 3A) and occur at positions of tively) with residues 99–314 of BN, the region comprising the two near-superposition in the models. These homology models should internal structural regions of BN. Also shown in Fig. 3A are the serve as effective working models for the folded structure of secondary structural elements observed in the BN solved struc- domains B and C, which will of course be supplanted by the actual ture. Fig. 3B shows the topological pattern of 9 ␣-helices and 13 crystal structure of E. coli CPS (29) as soon as possible. ␤-strands shared by BN and the other members of the N-ligase Site-Directed Mutagenic Analysis of CPS. We have carried out structural group (ref. 26). site-directed mutagenesis studies to determine whether appro- The ATP binding sites have been determined to be in equiv- priate CPS residues display the functional characteristics ex- alent positions for GS (27) and AA (28), and are suggested to be pected for ATP utilization via the Walker A͞B structural motif in corresponding positions for BN (24, 26; Fig. 3B) and E. coli and͞or for ATP utilization via a palmate structural motif. For CPS (29). This ATP site is not comprised of the frequently domain B, CPS peptide 227–242 (KSLKGWKEVEYEVVRD) observed ␤␣␤ nucleotide binding fold, where (i) a glycine-rich conforms most closely (6) to the primary sequence expected for Walker A sequence forms a loop (the P-loop) about the nucle- the Walker B motif (K͞RXXXGXXXLOOO-space-D, where X otide triphosphate group, with the loop connecting a ␤ strand can be any residue and O is a generally nonpolar with an ␣ helix, and (ii) a Walker B sequence corresponds to a residue; ref. 4). In addition, CPS 227–242 is the counterpart of the hydrophobic strand of parallel ␤ sheet, terminated by a carbox- rat liver CPS peptide identified as near the ATPB site by affinity ylate, that flanks the triphosphate binding site (4). Instead, analog labeling (12). We have previously shown that Y237 and interaction with ATP involves a ‘‘palmate’’ motif (27) comprised R241 are not critical residues, but that E234, E236, and E238 are of two structural components: (i) a 4-stranded antiparallel ␤- critical for function and that D242 appears to occupy a critical sheet (␤-strands 5–8 in Fig. 3B) plus the loop connecting strands structural position (33). These findings were not consistent with 6 and 7, and (ii) a 5-stranded antiparallel ␤-sheet (␤-strands 9–13 the expectations for conventional functioning of a Walker B motif in Fig. 3B) plus the loop connecting strands 10 and 11. The ATP that D242 would be critical for function and that hydrophobic is sandwiched between these two ␤-sheets, binding at the face of interaction would constitute the sole functional role of residues the sheets. Both loops of this motif are flexible, presumably to 236–241. In the present study, we have created mutations tar- permit entry of ATP and substrates and then to swing inward to geting four additional residues within the CPS peptide 227–242 cover and stabilize the substrates and intermediates during ca- that are conserved in all or most CPSs: K227A, G231V, W232A, talysis. The loops also serve to exclude water from the large and K233A. We were especially interested in the behavior of catalytic cavity. In addition to the loop mobility, the second region mutants involving K227 and G231 because these residues would of BN (␤5–␤8) appears to serve as a ‘‘lid’’ that closes down on the be expected to be critical if CPS utilizes the standard interactions when the substrates are positioned for catalysis. with ATP previously observed for the Walker B sequence motif. We utilized the x-ray coordinates for residues 99–314 of BN (24) The functional effect of each mutation was determined by in vivo as a template for modeling the homologous portions of CPS screening with the CPS-disrupted yeast strain LPL26 (41). All domains B and C (residues 133–343 and 677–887, respectively), four mutants displayed wild-type functional behavior, indicating

FIG.3. (A) Alignment of yeast arginine-specific CPS domain B (residues 133–343) and domain C (residues 677–887) with BN (residues 99–314; ref. 42). The alignment is essentially that of Simmer et al. (2) and is based on alignment of five different CPSs. Vertical lines indicate sequence identity between the CPS domains or between BN and one or both of the CPS domains; the identical residues are shown in bold. a, b, and t indicate ␣-helix, ␤-sheet, and turn structural elements, respectively, in the solved structure of BN, and ? indicates residues too disordered to be included in the solved structure (24). The 12 BN residues conserved or conservatively substituted in the ATP binding sites of BN, GS, and AA (26) are indicated by underlining of the BN residues and by stars above their positions. (B) Topological diagram of residues 1–314 of the BN structure, simplified from Artymiuk et al. (26). ␣-Helix elements are shown as circles and ␤-sheet elements as arrows. The unshaded elements (helices A–I and sheets 1–13; also indicated on the bottom line of A) are those that are found in all members of the N-ligase family and the shaded elements are those unique to BN (26). The topology of residues 315–449 (data not shown) is also unique to BN. Downloaded by guest on September 25, 2021 Biochemistry: Kothe et al. Proc. Natl. Acad. Sci. USA 94 (1997) 12351

Walker A motif, supported growth at Ϸ30% of the wild-type rate. These findings, together with those on the potential Walker B motifs of domains B and C, very strongly suggest that CPS is not a member of the Walker A͞B motif structural family. It should be noted that another region of CPS (including G196 and G199 of domain B and G741 of domain C) has been proposed as the Walker A motif, and that these corresponding three glycyl residues were critical for activity of E. coli CPS (22). However, this domain C region of CPS contains only one glycyl residue and neither the domain B nor the domain C region contains the expected terminal K or S͞T residues; also, there is no correctly spaced Walker B motif for this other potential Walker A motif. Our site-directed mutagenic analysis provides support for utilization of a ‘‘palmate’’ ATP binding motif by CPS and argues against its utilization of a Walker A͞B ATP binding motif. The glycine-containing loop in the palmate motif (between ␤6 and ␤7) differs from that of the Walker A motif by not containing the terminal conserved residues KS͞T (27). The CPS region con- taining G196 and G199 (domain B) or G741 (domain C) form this loop in our modeled structure, whereas the other CPS region considered as a potential Walker A motif (131–139 of domain B or 675–683 of domain C) forms a loop that connects the ATP-binding region of CPS to another region of the protein. In the palmate structure, the homologous CPS peptides 227–242 (domain B) and 773–788 (domain C) form ␤-strand 9, which contains 1 of the 12 conserved active site residues: the critical residues E234 and E780 are thus predicted to form hydrogen bonds with the ribose hydroxyl group of ATP. To further confirm that the palmate motif modeling is correct, we carried out mutagenic screening of CPS residues E317 and N319, which would correspond to 2 more of the 12 residues conserved at the ATP binding sites of known palmate motif enzymes (26). Both E317 and N319 are expected to be near the ATP phosphate groups and to be coordinated to the Mg2ϩ.In FIG. 4. Superimposed C␣ traces of BN residues 99–314 (dark) with keeping with these proposed critical roles, neither E317A nor residues 133–343 of CPS domain B (Upper) or residues 677–887 of CPS domain C (Lower). Table 1. Effect of CPS mutations on the ability of transformants to support the growth of the CPS-disrupted strain LPL26 and on that none of the targeted residues plays a critical role in CPS CP synthesis activity in crude extracts prepared from transformants. structure and͞or function. Generation We have also carried out site-directed mutagenic screening of time, (h) Specific activity the domain C equivalent of the putative Walker B motif: peptide 773–784 (KFIEGAQEIDVD). We have targeted the residues Mutant 30°C 37°C Ϫ Gly ϩ Gly corresponding to those that were found to be critical within Wild-type CPS 4.9 5.7 1.05 3.35 domain B: E780, D782, and D784. Mutations were designed to (i) determine if any feature of the side chain is essential (E780A, K773A 5.5 4.7 1.23 5.12 D782A, D784A); (ii) determine whether the spatial arrangement E780A u u u u of the carboxyl group is critical (E780D, D782E, D784E); (iii) E780D u u u u determine if hydrogen-bonding potential is sufficient for func- E780Q u u u u tionality or if the carboxyl groups are critical (E780Q, D782N, E780H u u u u D784N); and (iv) determine if a substitution of the corresponding residue in the BN sequence can allow functioning (E780H, D782A u u u u D782E, D784N). The analysis of these mutants (Table 1) revealed D782E 4.7 6.0 0.62 2.12 that the presence of a carboxyl group and its exact spatial D782N 33.5 96.3 0.16 0.26 arrangement appears to be critical only for E780, although D784A u u u u functional groups with hydrogen bonding potential are required D784E 4.8 4.8 0.27 0.57 at positions 782 and 784. We have also constructed and analyzed D784N u* u* 0.19 0.46 the K773A mutant (Table 1). Although the terminal lysyl residue D784Q 43.3 91.5 u u is a critical feature of the Walker B motif, its replacement by alanine resulted in an apparently fully functional CPS. E317A u u u u Site-directed mutagenic screening has also been carried out on N319A u u u u peptide 131–139 (KYNVKVLGT), which contains elements of Glycine activation indicates conformational coupling between the the Walker A sequence (K͞R-space-GXXXXGKT͞S) and oc- synthetase and glutamine amidotransferase subunits of CPS, with curs at the expected distance from the putative Walker B glycine binding to the latter as a glutamine analog (44); the presence sequence (6). When we substituted G138, which should be of such coupling serves as one indication that the mutation in CPS has not resulted in a gross conformational change. Specific activity is nmol essential if it is involved in forming a P loop, with the bulky amino CP synthesized/min per unit CPS protein, with amount of CPS protein acid valine, the construction displayed wild type functional be- determined by densitometry of Western blots. u, Below the assay havior in the in vivo screen. The T139A mutant, with elimination detection limit; u*, D784N failed to grow in liquid media but showed of the terminal threonine residue that is also critical for the essentially wild-type growth on solid medium. Downloaded by guest on September 25, 2021 12352 Biochemistry: Kothe et al. Proc. Natl. Acad. Sci. USA 94 (1997)

N319A displayed activity in the in vivo screen. The domain C carbamate as an intermediate (Eq. 3), presumably because it is in Ϫ counterpart to E317 is E861, which is conserved in all CPSs. equilibrium with the CPS substrates NH3 and HCO3 (31) and Previous studies have shown that the E. coli CPS analog of E861 would thus be very difficult to definitively identify. Two lines of is an essential residue (45). Thus, from the present and previous indirect evidence have suggested that ATPC functions in a studies, 4 of the 12 conserved palmate motif active site residues reaction (Eq. 4). The ability of domain C to are known to also be critical for CPS functioning. During the catalyze the formation of ATP from CP and ADP has been present studies, additional site-directed mutagenic analysis of E. interpreted as reflecting the reverse of Eq. 4 (47). However, as coli CPS was reported (32, 46). Significant functional changes previously noted (16), this reaction more likely reflects the resulted (32, 46) from the site-directed mutation of 3 more of the reversal of Eq. 1, CxP formation, with CP acting as a CxP analog 12 conserved palmate motif residues within both domains B and because several biotin-dependent enzymes (for which CxP is an C (counterparts to R150͞R694, R190͞R734, and Q303͞Q848) intermediate whereas neither carbamate nor CP is involved in the but not from the mutation of a fourth conserved residue in either reaction) can also catalyze the formation of ATP from CP plus domain (counterparts to E226͞S772). Consistent with our ADP (25). Furthermore, the synthesis of ATP from CP plus ADP present findings, the E. coli CPS counterparts of E234, E317, can also be catalyzed by (48) and formyltet- N319, and E780 were found to be critical (32); the domain C rahydrofolate synthetase (49), where presumably CP is serving, counterparts of E317 and N319 were also found to be critical (46). respectively, as an analog of glutamyl-phosphate or of formyl- In summary, both our own mutagenic analysis and the recently phosphate. Reported sequence identity (13.5%) between domain reported one on E. coli CPS are consistent with the proposal that CofE. coli CPS and Pseudomonas aerouginosa carbamate kinase domains B and C utilize a palmate ATP-binding motif. (50) has also been interpreted as supporting the idea of a Novel Mechanism for Coupled ATP Use by CPS: ATP Binding functional relationship between the two enzymes. However, the at One Domain Triggers a Conformational Switch That Allows sequence comparison (50) failed to consider the requirement for CP Synthesis at the Other Domain. The finding that domains B 22 gaps in the alignment to produce the observed 42 identities. and C are structurally equivalent and that they conform to the When re-analyzed with appropriate gap penalties (51), there is no structure expected for N- (our present work and ref. 29), as significant identity between carbamate kinase and either domain well as the recent finding that domains B and C are functionally B or C of CPS. equivalent (30), have prompted us to re-evaluate the sequential We are proposing an alternative mechanism (Fig. 2) for CPS reaction pathway for CPS (Fig. 1) that was proposed in 1960 by in which the energy derived from the 2 ATP molecules is coupled. Jones and Lipmann (31). Although domain B catalysis of the In Fig. 2, ATPC binds to domain C and acts as a nucleotide switch carbamate synthesis reaction described by Eqs. 1-3 is consistent for domain B, similar to ATP usage by , myosin, and with the known N-ligase functionality of the palmate structural signal transduction proteins (34). The ‘‘switched-on’’ conforma- Ϫ group, domain C catalysis of the carbamate kinase reaction of Eq. tion of domain B then sequentially binds ATPB, HCO3 , and NH3 4 is not. Nor is the catalysis of distinct reaction types consistent and carries out the portion of the N-ligase reaction described by with the structural equivalence and the apparent functional Eqs. 1 and 2. However, rather than proceeding through Eq. 3, the equivalence of domains B and C. Finally, utilization of the ‘‘switched-on’’ conformation of domain B allows the direct col- sequential pathway by independent active sites of homologous lapse of the tetrahedral intermediate to CP, as shown in Eq. 3૾. domains B and C would require movement of the labile carbam- This route of collapse is possibly facilitated by protonation of one ate intermediate from domain B to domain C, an apparent of the hydroxyl groups of the tetrahedral intermediate, yielding distance of 35 Å (29). water as the actual leaving group. Because ADPB now occupies It is very well established (14–23) that one molecule of ATP the ATP͞ADP site of domain B, it cannot act as a nucleotide binds to domain B and is used to form CxP (Eq. 1), but the validity switch for domain C. Thus, domain C cannot attain the of Eqs. 2-4 has only been assumed by CPS researchers (1, 30, 32, switched-on conformation necessary for Eq. 3૾. Instead, a stan- 33, 45). There have been no reported attempts to directly identify dard N-ligase reaction yielding ADPC, Pi, and carbamate occurs

Table 2. Consistency of experimental evidence with the proposed mechanisms for CPSase Consistent with proposed mechanisms Experimental evidence Sequential model (Fig. 1) Coupled model (Fig. 2) 1. Functional equivalence of domains B and C (30) No Yes 2. Catalysis of -dependent cleavage of ATP to ADP ϩ Pi (47) Yes Yes 3. Domain B localization of bicarbonate-dependent ATPase activity (22, 23) Yes Yes 4. Catalysis of ATP formation from CP ϩ ADP (47) Yes Yes 5. Domain C localization of ATP formation from CP ϩ ADP (22, 23) Yes Yes 6. Significant increases in the Km for ATPB observed in ATPC site mutants No Yes but not in the Km for ATPC in response to corresponding mutations at the ATPB site (22, 32, 46) 7. Structural equivalence of domains B and C to each other and to the No Yes N-ligases (29) 8. Absence of significant sequence identity for domain B or C with No Yes carbamate kinase 9. Expected lability of carbamate during movement from domain B to No Yes domain C occurring in the sequential mechanism 10. Chemical and kinetic competency of carboxy-phosphate (14–21) Yes Yes 11. Absence of ATP͞ADP or ATP/Pi exchange, indicating ADP remains Yes Yes bound as a component of the enzyme–carboxy–phosphate complex (18–20) 12. ATP positional isotope exchange evidence for an alternative pathway No Yes forming carboxy-phosphate in addition to the main CP-forming pathway (18, 21) 13. Binding of both molecules of ATP before ammonia is bound (16, 52–54) No Yes Downloaded by guest on September 25, 2021 Biochemistry: Kothe et al. Proc. Natl. Acad. Sci. USA 94 (1997) 12353

on domain C (Eqs. 1-3), and triggers a ‘‘switched-off’’ confor- anism and of a solved three-dimensional structure for CPS should mation that allows release of the products bound to both domains greatly facilitate the rational design and interpretation of future (CP, Pi, carbamate, and 2 ADP). Upon dissociation from the experiments, and yield significant insights into the CPS structure͞ Ϫ enzyme, carbamate would form NH3 and HCO3 , the substrates function relationship. for the overall reaction. For the CPSs where NH3 is delivered to domain B from the domain (rather than free NH We gratefully acknowledge Dr. Angela L. Lim for her helpful advice 3 and Dr. Richard Deth for his assistance with homology modeling. binding directly), it is likely that NH3 will not also be delivered to These studies were supported by research grants from the National domain C and that water will instead serve to break down the CxP Ϫ Institutes of Health (DK-33460) and from the Northeastern University intermediate, yielding HCO3 and phosphate. Research and Scholarship Development Fund. As outlined in Table 2, our proposed coupled scheme recon- 1. Anderson, P. M. (1995) in and Excretion, eds. Walsh, P. J. ciles the many previous experimental findings on the CPS mech- & Wright, P. (CRC, Boca Raton, FL), pp. 33–49. anism, whereas a number of the previous findings are not 2. Simmer, J. P., Kelly, R. E., Rinker, A. G., Jr., Scully, J. L. & Evans, D. R. (1990) consistent with the sequential scheme. The coupled scheme J. Biol. Chem. 265, 10395–10402. 3. van den Hoff, M. J. B., Jonker, A., Beintema, J. J., Lamers, W. H. (1995) J. Mol. predicts functionally equivalent domains (point 1) because the Evol. 41, 813–832. distinction between the functions carried out by domains B and 4. Walker, J. E., Saraste, M., Runswick, M. J. & Gay, N. J. (1982) EMBO J. 1, 945–951. 5. Lusty, C. J., Widgren, E. E., Broglie, K. E. & Nyuonya, H. (1983) J. Biol. Chem. C is determined only by which of these domains first binds ATP. 258, 14466–14472. Our present assignment of dedicated functions for domains B and 6. Powers-Lee, S. G. & Corina, K. (1987) J. Biol. Chem. 262, 9052–9056. C within intact CPS was based on the domain localizations of 7. Powers-Lee, S. G. & Corina, K. (1986) J. Biol. Chem. 261, 15349–15352. 8. Marshall, M. A. & Fahien, L. A. (1988) Arch. Biochem. Biophys. 262, 455–470. partial CPS activities (points 2–5), but is further supported by the 9. Evans, D. R. & Balon, M. A. (1988) Biochim. Biophys. Acta 953, 185–196. recent finding that the glutaminase domain of E. coli CPS is near 10. Mareya, S. M. & Raushel, F. M. (1995) Bioorg. Med. Chem. 3, 525–532. 11. Kim, H., Lee, L. & Evans, D. R. (1991) Biochemistry 30, 10322–10329. domain B and removed from domain C (29). The existence of the 12. Potter, M. D. & Powers-Lee, S. G. (1992) J. Biol. Chem. 267, 2023–2031. two partial reactions and their domain localizations (points 2–5) 13. Potter, M. D. & Powers-Lee, S. G. (1993) Arch. Biochem. Biophys. 306, 377–382. are consistent with both proposed schemes. In both schemes, the 14. Anderson, P. M. & Meister, A. (1965) Biochemistry 4, 2803–2809. Ϫ 15. Powers, S. G. & Meister, A. (1976) Proc. Natl. Acad. Sci. USA 73, 3020–3024. HCO3 -dependent ATPase activity is thought to reflect CxP 16. Rubio, V. & Grisolia, S. (1977) Biochemistry 16, 321–329. formation at domain B. The ability of CPS to catalyze the 17. Powers, S. G. & Meister, A. (1978) J. Biol. Chem. 253, 1258–1265. 18. Wimmer, M. J., Rose, I. A., Powers, S. G. & Meister, A. (1979) J. Biol. Chem. 254, formation of ATP from CP and ADP is interpreted as reflecting 1854–1859. the reverse of Eq. 4 (47) in the sequential mechanism. In the 19. Raushel, F. M. & Villafranca, J. J. (1979) Biochemistry 18, 3424–3429. coupled scheme it is interpreted as the reversal of Eq. 1, CxP 20. Rubio, V., Britton, H. G., Grisolia, S., Sproat, B. S. & Lowe, G. (1981) Biochemistry 20, 1969–1973. formation, with CP acting as a CxP analog (16) and with the ADP 21. Meek, T. D., Karsten, W. E. & DeBrosse, C. W. (1987) Biochemistry 26, binding at domain C, because this is the site for initial ATP 2584–2593. 22. Post, L. E., Post, D. J. & Raushel, F. M. (1990) J. Biol. Chem. 265, 7742–7747. binding. However, only the coupled scheme can provide a ratio- 23. Alonso, E., Cervera, J., Garcia-Espana, A., Bendala, E. & Rubio, V. (1992) J. Biol. nale for the finding (point 6) that many of the site-directed Chem. 267, 4524–4532. 24. Waldrop, G. L., Rayment, I. & Holden, H. M. (1994) Biochemistry 33, 10249–10256. mutations at the ATPC site have a significant effect on the Km for Ϫ 25. Knowles, J. R. (1989) Annu. Rev. Biochem. 58, 195–221. ATPB (measured in the assay for HCO3 -dependent ATPase 26. Artymiuk, P. J., Poirrette, A. R., Rice, D. W. & Willett, P. (1996) Nat. Struct. Biol. 3, 128–132. activity) whereas the corresponding ATPB site mutations fail to 27. Yamaguchi, H., Kato, H., Hata, Y., Nishioka, T., Kimura, A., Oda, A. & Katsube, appreciably affect the Km for ATPC (measured in the assay for Y. (1993) J. Mol. Biol. 229, 1083–1100. ATP synthesis from ADP plus CP). These kinetic effects would be 28. Fan, C., Moew, P. C., Walsh, C. T. & Knox, J. R. (1994) Science 266, 439–443. 29. Thoden, J. B., Holden, H. M., Wesenberg, G., Raushel, F. M. & Rayment, I. consistent with the required binding of ATPC before ATPB in Fig. (1997) Biochemistry 36, 6305–6316. 2 but not with the independent and sequential utilization of the two 30. Guy, H. I. & Evans, D. R. (1996) J. Biol. Chem. 271, 13762–13769. ATP molecules previously proposed in Fig. 1. The previous 31. Jones, M. E. & Lipmann, F. (1960) Proc. Natl. Acad. Sci. USA 46, 1194–1205. 32. Stapleton, M. A., Javid-Majd, F., Harmon, M. F., Hanks, B. A., Grahmann, J. L., observation (55, 56) that allosteric control of CPS is centered on Mullins, L. S. & Raushel, F. M. (1996) Biochemistry 35, 14352–14361. binding of ATPC would also be consistent with the coupled scheme 33. Zheng, W., Lim, A. L. & Powers-Lee, S. G. (1997) Biochim. Biophys. Acta 1341, 35–48. 34. Schindelin, H., Kisker, C., Schlessman, J. L., Howard, J. B. & Rees, D. C. (1997) where ATPC binds first in the intact CPS. Nature (London) 387, 370–376. As discussed above, points 7–9 of Table 2 fit only the coupled 35. Ruvinov, S. B., Yang, X. J., Parris, K. D., Banik, U., Ahmed, S. A., Miles, E. W. scheme. Generally the experimental evidence supporting the & Sackett, D. L. (1995) J. Biol. Chem. 270, 6357–6369. 36. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) Molecular Cloning: A Laboratory intermediacy of CxP is consistent with both proposed mecha- Manual (Cold Spring Harbor Lab. Press, Plainview, NY), 2nd Ed. nisms because they both involve its formation (points 10 and 11). 37. Rose, M. D., Winston, F. & Hieter, P. (1990) Methods in Yeast Genetics (Cold Spring Harbor Lab. Press, Plainview, NY). However, the isotopic exchange evidence for a pathway forming 38. Higuchi, R. (1990) in PCR Protocols, eds. Innis, M. A., Gelfand, D. H., Sninsky, CxP which is distinct from its formation via the main CP-forming J. J. & White, T. J. (Academic, New York), pp. 177–183. pathway (point 12) is only consistent with the coupled scheme. 39. Deng, W. P. & Nickoloff, J. A. (1992) Anal. Biochem. 200, 81–88. 40. Lim, A. L. & Powers-Lee, S. G. (1997) Arch. Biochem. Biophys. 339, 344–352. The finding in a number of studies that both molecules of ATP 41. Lim, A. L. & Powers-Lee, S. G. (1996) J. Biol. Chem. 271, 11400–11409. bind before NH3 (point 13) is also consistent only with the 42. Kondo, H., Shiratsuhi, K., Yoshimoto, T., Masuda, T., Kitazono, A., Tsuru, D., Anai, M., Skiguchi, M. & Tanabe, T. (1991) Proc. Natl. Acad. Sci. USA 88, 9730–9733. coupled scheme. This order of binding is also predicted by one 43. Brooks, B. R., Bruccoleri, R. E., Olafson, B. D., States, D. J., Swaminathan, S. kinetic analysis (57), although two others (21, 58) find that the & Karplus, M. (1983) J. Comput. Chem. 4, 187–217. best fit to their data is given by NH binding after the first ATP. 44. Pierard, A. & Schroter, B. (1978) J. Bacteriol. 134, 167–176. 3 45. Guillou, F., Liao, M., Garcia-Espana, A. & Lusty, C. J. (1992) Biochemistry 31, Rubio and Grisolia (16) have previously noted the inconsistency 1656–1664. of the direct binding studies with the sequential mechanism and 46. Javid-Majd, F., Stapleton, M. A., Harmon, M. F., Hanks, B. A., Mullins, L. S. & Raushel, F. M. (1996) Biochemistry 35, 14362–14369. suggested a concerted reaction between CxP, NH3, and the 47. Anderson, P. M. & Meister, A. (1966) Biochemistry 5, 3157–3163. second molecule of ATP to form CP with no intermediate 48. Tate, S. S., Leu, F. Y. & Meister, A. (1972) J. Biol. Chem. 247, 5312–5321. formation of carbamate. This mechanism, however, would not be 49. Buttlaire, D. H., Himes, R. H. & Reed, G. H. (1976) J. Biol. Chem. 251, 4159–4161. consistent with points 1, 6, 7, or 13 of Table 2. An additional 50. Baur, H., Luethi, E., Stalon, V., Mercenier, A. & Haas, D. (1989) Eur. J. Biochem. alternative CPS mechanism has also been considered (30), where 179, 53–60. 51. Feng, D. F., Johnson, M. S. & Doolittle, R. F. (1984) J. Mol. Evol. 21, 112–125. each domain can catalyze the overall reaction by the sequential 52. Fahien, L. A. & Cohen, P. P. (1964) J. Biol. Chem. 239, 1925–1934. mechanism but only when associated with another domain in a 53. Powers, S. G. & Meister, A. (1978) J. Biol. Chem. 253, 800–803. pseudodimer. However, this mechanism would be inconsistent 54. Rubio, V., Britton, H. G. & Grisolia, S. (1979) Eur. J. Biochem. 93, 245–256. 55. Liu, X., Guy, H. I. & Evans, D. R. (1994) J. Biol. Chem. 269, 27747–27755. with points 3, 5, 6, and 7 of Table 2. It thus seems that the existing 56. Czerwinski, R. M., Mareya, S. M. & Raushel, F. M. (1995) Biochemistry 34, experimental evidence supports our present proposal that CPS 13920–13927. 57. Elliott, K. R. F. & Tipton, K. F. (1974) Biochem. J. 141, 817–824. functions via coupled usage of the two molecules of ATP as 58. Raushel, F. M., Anderson, P. M. & Villafranca, J. J. (1978) Biochemistry 17, depicted in Fig. 2. The availability of this novel proposed mech- 5587–5591. Downloaded by guest on September 25, 2021